International Journal of Obesity (1999) 23, 1276±1281
ß 1999 Stockton Press All rights reserved 0307±0565/99 $15.00
http://www.stockton-press.co.uk/ijo
Body composition of healthy 7- and 8-year-old
children and a comparison with the `reference
child'
CHS Ruxton1*, JJ Reilly2 and TR Kirk3
1
3
The Sugar Bureau, Dolphin Square, London, UK; 2University Department of Human Nutrition, Yorkhill Hospitals, Glasgow, UK;
Centre for Food Research, Queen Margaret College, Edinburgh, UK
BACKGROUND: There are few longitudinal data on body composition in healthy children. This has prompted a
reliance on notional standards such as the `reference child', to validate new methods of determining body composition and comparing cross-sectional height, weight and fatness data.
OBJECTIVES: These were twofold Ð to provide normative longitudinal data on changes in body composition in
healthy pre-pubertal children, and to compare measures of growth and body composition with the appropriate agespeci®c reference child.
DESIGN: A sample of healthy Scottish children aged 7 ± 8 y (n ˆ 257) was recruited during 1991=1992. Data on height,
weight, skinfold thickness and resistance from bioelectrical impedance analysis were collected twice, 12 months
apart. Percentage body fat was estimated from both skinfolds and bioelectrical impedance.
RESULTS: Fat and fat-free mass, but not body mass index, differed between boys and girls. All measurements
increased signi®cantly over the 12 month period except percentage body fat from skinfolds in boys. The reference
child comparison revealed that our sample was taller, heavier and fatter and gained weight and fat mass at a greater
rate than the Fomon standards.
CONCLUSIONS: Data from the children in this study suggest that the reference child has a body composition which is
now out of date. This may have important implications for body composition methodology. New references for height
and weight may be required, but an upgrading of the body fat reference may con¯ict with public health aims to reduce
obesity.
Keywords: body composition; standards; children
Introduction
Measurement of body composition in childhood is of
considerable importance and has a variety of applications,1,2 such as assessment of growth and nutritional
status, public health (in de®ning the prevalence of
obesity) and interpreting energy expenditure data. Our
current understanding of the normal development of
body composition during childhood is heavily dependent on the concept of Fomon's `reference child'.3
This gives estimates of normal body composition
based on a combination of theoretical considerations
and empirical data on height and weight collected
from children in the USA many years ago.4 The
reference child concept represents notional values
for normative fat and fat-free mass (FFM) at particular
ages in childhood. These in turn give rise to assumptions concerning hydration, density and potassium
content of FFM which underpin many of the historical
*Correspondence: Dr C Ruxton, Research Manager, The Sugar
Bureau, Duncan House, Dolphin Square, London SW1A 3PW, UK.
E-mail: [email protected]
Received 7 January 1999; revised 25 May 1999; accepted 22 June
1999
and current methods for measuring body composition.
While the concept of the reference child has been
extremely useful in the development of age- and sexspeci®c constants in body composition, and in the
interpretation of body composition measurements,
there are two major concerns about the existing data
used to derive the `reference child'. First, secular
trends in body fatness are occurring rapidly in the
developed world in both children5,6 and adults.7
Second, the authors of the original reference data
noted that their normative data should be considered
preliminary and crude because of uncertainties about
the data and the large number of assumptions
required. Empirical data on body composition of
normal healthy children are scarce, and empirical
data on actual longitudinal changes in body composition during childhood are seldom available.
The aims of this study were to: (a) obtain normative
data on body composition and changes in body
composition measured longitudinally in a sample of
healthy Scottish 7- and 8-y-old children; and (b)
compare these with data on the reference child. In
addition to the estimates from skinfolds, data on fat
mass, FFM and body fat from bioelectrical impedance
analysis (BIA) are also presented since there are few
longitudinal data collected by this method in the
literature.
Body composition in children
CHS Ruxton et al
Methods
Subjects
Following ethical approval from Queen Margaret
College and Lothian Education Authority, healthy
children aged 7 ± 8 y attending ®ve primary schools
in Lothian Region, Scotland, were recruited for a
study of dietary intake and anthropometry. Data
collection for the study ran from February 1991 to
March 1992, excluding school holidays. The children
were a representative group for Edinburgh children of
this age in terms of social class and gender, and have
been described in detail elsewhere.8 Height, weight
and skinfold thickness were measured during school
hours in 98% of the target group (n ˆ 257). Eleven
children were not measured due to absenteeism on the
measurement day, and one child refused to be measured. After approximately 12 months, the measurements were repeated, this time obtaining 90% of the
original sample (n ˆ 240). The 29 children not measured at the follow-up included 13 absentees, one
child who refused (same child as on the ®rst occasion), and 15 children who had moved to other
schools.
Height
A portable stadiometer (Minimetre, Child Growth
Foundation, UK) measuring to 180 cm in 0.1 cm
increments was used to measure height after being
checked for accuracy in its position on the wall.
Children were positioned and height was measured
to the nearest 0.1 cm according to the method of
Tanner et al.9
Weight
A set of ¯oor scales (Soenle, Germany) weighing to
127 kg in 0.5 kg increments was used to measure
weight after being checked for accuracy using a set
of standard weights, weighing in 0.5 kg increments to
55 kg. Weight to the nearest 0.5 kg was measured after
asking the child (wearing light indoor clothing and
with shoes removed) to stand motionless on the scales
with feet together. Body mass index (BMI) was
computed as weight (kg)=height2(m).
Skinfold thickness
A set of metal Harpenden skinfold callipers was used
(British Indicators Ltd, UK), which exert a constant
pressure of 10 g=mm, measuring to 40 mm in 0.1 mm
increments. The callipers were calibrated by the
manufacturers prior to being used in the study, after
which they were not used by any person other than the
observer. Skinfold thicknesses at the tricep and subscapula regions were located and measured according
to the technique described by Tanner et al.8 This was
repeated three times on the left-hand side of the body
and the mean of the measurements recorded to the
nearest 1 mm. Fat mass, FFM and percentage body
fat were calculated using the equation of Slaughter
et al,10 which has been shown to produce unbiased
estimates of fatness validated against hydrodensitometry in Scottish pre-pubertal children.11 All skinfold
measurements were made by a single trained observer
(CHSR).
Bioelectrical impedance analysis (BIA)
The equipment used was a BIA101 plethysmograph
(RJL Systems, USA), measuring resistance to 1000 O
in 1 O increments. BIA data were taken from 135
(52%) children at the ®rst measurement and 238
(93%) children at follow-up under unfasted conditions. The lower rate of participation at the ®rst
measurement was due to a delay in ethical approval
for this part of the study and affected data collection
in two schools. Measurements of impedance were
taken with the child placed in the supine position
with legs slightly apart and arms not touching the
body. A constant current of 80 mA at a signal frequency of 50 kHz was then applied. The equation of
Houtkooper et al 12 was used to calculate FFM since it
has been successfully cross-validated against hydrodensitometry in pre-pubertal Scottish children.13 This
equation is as follows:
FFM …kg† ˆ 0:61 …RI† ‡ 0:25 …weight; kg† ‡ 1:31
where RI equals height2 (cm)=resistance (ohms).
Fat mass (kg) was then calculated by subtracting
FFM from weight and both were expressed as a
proportion of weight. Percentage body fat values
which fell below a selected cut-off point of 3.99%
were deemed to be unphysiological1 and were
excluded from subsequent analyses. This affected
®ve cases at baseline and two at follow-up and may
have been due to the low body weight and height of
these children producing false results when resistance
was measured.
Statistical analysis
Data were entered into SPSS 7.5 for Windows 95 and
comparisons were made between baseline and followup values using paired t-tests. In the case of estimates
of fat and FFM from BIA, this resulted in a drop in
sample size due to the smaller group participating at
baseline. Differences between boys and girls were
examined using independent t-tests. Signi®cance was
de®ned as P < 0.05. A comparison was made with the
reference child by calculating con®dence intervals for
each anthropometric variable. Our sample were
deemed to be statistically different if the reference
child value lay outside the appropriate con®dence
interval.
1277
Body composition in children
CHS Ruxton et al
1278
Results
The mean age of the boys was 7.49 (0.37) y at baseline
and 8.51 (0.39) y at follow-up, while the corresponding mean ages for girls were 7.52 (0.36) and 8.53
(0.39) y.
Mean (s.d.) values for height, weight, BMI and
measures of fatness calculated from skinfold thickness
are given in Table 1 alongside values for the reference
child. There were no signi®cant gender differences for
height, weight and BMI at either age. However, fat
mass in girls was greater at age 8 (P < 0.05), while
FFM was greater in boys at ages 7 and 8 (P < 0.001).
Table 1
In comparison with the reference child, boys and
girls were signi®cantly taller, heavier, fatter and had a
greater FFM at both ages. However, when percentage
FFM was considered, boys and girls had lower levels
than the reference child. The con®dence intervals for
each of the anthropometric and body composition
variables are given in Table 1.
Measures of fatness calculated from BIA are given
in Table 2. FFM and percentage FFM were greater in
boys than girls at age 7 (P < 0.05) and at age 8
(P < 0.005). Fat mass and percentage body fat were
lower in boys compared with girls at age 7 (P < 0.01)
and at age 8 (P < 0.001).
Table 3 shows how height, weight and measures of
fatness changed over the 12 months in boys and girls.
Comparison of absolute values with Fomon reference data
Age 7
Variables
Reference
Boys
Height (cm)
121.7
Weight (kg)
22.9
BMI
15.5
Fat mass (kg)
2.9
Percentage body fat
12.8
FFM (kg)
19.9
Percentage FFM
87.2
Girls
Height (cm)
120.6
Weight (kg)
21.8
BMI
15.0
Fat mass (kg)
Observed (s.d.) CI
16.8
FFM (kg)
18.1
Percentage FFM
83.2
Reference
n ˆ 133
125.0 (5.4)
124.1 ± 125.9
25.1 (4.1)
24.4 ± 25.8
16.0 (1.8)
15.7 ± 16.4
4.4 (2.6)
3.9 ± 4.8
16.5 (6.1)
15.5 ± 17.6
20.7 (2.5)
20.3 ± 21.2
83.4 (6.0)
82.4 ± 84.5
n ˆ 124
124.0 (5.2)
123.1 ± 124.9
24.2 (4.2)
23.6 ± 25.1
15.8 (2.0)
15.4 ± 16.1
4.6 (2.5)
4.4 ± 5.3
19.0 (6.4)
17.8 ± 20.1
19.6 (2.5)
19.1 ± 20.0
81.0 (6.4)
79.9 ± 82.1
3.7
Percentage body fat
Age 8
Observed (s.d.) CI
n ˆ124
130.2 (5.7)
129.2 ± 131.3
28.2 (4.6)
27.3 ± 29.0
16.5 (2.0)
16.2 ± 16.9
4.7 (2.8)
4.4 ± 5.4
16.9 (6.4)
15.8 ± 18.0
23.5 (2.8)
22.7 ± 23.7
83.1 (6.4)
81.9 ± 84.2
n ˆ 116
130.0 (5.5)
128.5 ± 130.6
27.8 (4.9)
26.9 ± 28.7
16.5 (2.2)
16.1 ± 16.9
5.8 (2.9)
5.3 ± 6.4
20.0 (6.3)
18.9 ± 21.2
22.0 (2.5)
21.5 ± 22.4
79.9 (6.2)
78.8 ± 81.1
127.0
25.3
15.7
3.3
13.0
22.0
87.0
126.4
24.8
15.5
4.3
17.4
20.5
82.6
Fat mass, FFM and percentage body fat estimated from skinfolds.10
CI ˆ con®dence interval.
Table 2
Fat mass, FFM and percentage body fat from BIA at ages 7 and 8 y
Age 7
Variable
Mean
Boys
Fat mass (kg)
Percentage body fat
FFM (kg)
Percentage FFM
Girls
Fat mass (kg)
Percentage body fat
FFM (kg)
Percentage FFM
n ˆ 65
3.4
13.3
21.0
86.7
n ˆ 75
4.0
16.0
20.0
84.0
CI ˆ con®dence interval.
Age 8
s.d.
CI
2.0
5.4
2.7
5.4
2.7 ± 3.8
11.5 ± 14.4
20.2 ± 21.6
85.4 ± 88.0
2.1
6.1
2.8
6.1
3.4 ± 4.4
14.3 ± 17.2
19.3 ± 20.6
82.6 ± 85.4
Mean
n ˆ 122
4.3
14.8
24.0
85.2
n ˆ 116
5.2
18.2
22.6
81.8
s.d.
CI
2.2
5.1
3.2
5.1
3.6 ± 4.8
13.5 ± 16.0
22.5 ± 24.0
84.3 ± 86.2
2.4
5.6
3.2
5.6
4.7 ± 5.9
17.2 ± 20.0
21.6 ± 23.1
80.8 ± 83.0
Body composition in children
CHS Ruxton et al
Table 3
1279
Changes in height, weight, BMI, FFM and percentage body fat over 12 months
Variable
Boys
Height (cm=y)
Weight (kg=y)
BMI
Percentage body fat (skinfold)
Percentage FFM (skinfold)
Fat mass (kg) BIA
Percentage body fat (BIA)
FFM (kg) BIA
Percentage FFM (BIA)
Girls
Height (cm=y)
Weight (kg=y)
BMI
Percentage body fat (skinfold)
Percentage FFM (skinfold)
Fat mass (kg) BIA
Percentage body fat BIA
FFM (kg) BIA
Percentage FFM (BIA)
n
Mean change
(s.d.) a
Significance bof
change; P <
Correlation between
baseline and follow up; r ˆ
120
120
120
120
120
56
56
56
56
5.4 (1.6)
3.2 (1.7)
0.54 (1.4)
0.6 (3.3)
7 0.6 (3.3)
1.0 (0.9)
1.9 (3.1)
0.1 (2.3)
7 1.8 (3.1)
0.001
0.001
0.001
±
0.05
0.001
0.001
0.001
0.001
0.86
0.82
0.95
0.75
0.86
0.90
0.96
0.94
0.82
116
116
116
116
116
68
68
68
68
5.6
3.4
0.7
1.2
7 1.2
1.4
2.8
2.3
7 2.8
0.001
0.001
0.001
0.001
0.001
0.001
0.001
0.001
0.001
0.84
0.85
0.83
0.98
0.84
0.85
0.87
0.96
0.85
(1.2)
(1.8)
(1.1)
(3.5)
(3.5)
(1.2)
(3.3)
(1.0)
(3.2)
a
For height and weight, decimal time between the baseline and follow-up measurements was taken into account.
Signi®cance of change analysed using a paired t-test.
b
Paired longitudinal data were available for 120 boys
and 116 girls for measures which did not relate to
BIA. All mean values were highly correlated and
increased signi®cantly over the 12 months, except
percentage FFM in both sexes, which signi®cantly
decreased, and percentage body fat from skinfolds in
boys, which remained constant. Height and weight
velocity were similar between the sexes. Girls
appeared to gain body fat and lose FFM at a greater
rate than boys but any apparent differences were not
statistically signi®cant.
Table 4 compares the changes in height, weight,
FFM and fat mass (both from skinfolds) with expected
changes in the reference child. Con®dence intervals
for the variables are also given. When compared with
the `expected' changes from Fomon's3 data, our boys
and girls gained both weight and fat mass at a
Table 4 Comparison with Fomon reference data Ð changes in
height (cm=y) and body composition (kg=y) for ages 7 ± 8 y
Variable
Reference
Observed (s.d.) CI
Boys n ˆ 120)
Change in height (cm)
‡ 5.1
Change in weight (kg)
‡ 2.4
Change in fat mass (kg)
‡ 0.4
Change in FFM (kg)
‡ 2.0
Girls (n ˆ 116)
Change in height (cm)
‡ 5.4 (0.14)
5.1 ± 5.6
‡ 3.2 (1.6)a
2.9 ± 3.5
‡ 0.7 (1.2)a
0.5 ± 1.0
‡ 2.5 (1.3)a
2.3 ± 2.7
‡ 5.8
Change in weight (kg)
‡ 2.9
Change in fat mass (kg)
‡ 0.6
Change in FFM (kg)
‡ 2.3
Fat mass and FFM calculated from skinfolds.10
a
Signi®cantly different from Fomon standard.
‡ 5.6 (0.11)
5.4 ± 5.8
‡ 3.4 (1.9)a
3.1 ± 3.8
‡ 1.0 (1.3)a
0.8 ± 1.3
‡ 2.4 (1.2)
2.1 ± 2.6
signi®cantly greater rate than the reference child.
The increase in absolute FFM was greater in boys
compared with the reference child but not in girls.
Gains in height were similar to the reference child in
both sexes.
Discussion
Despite similarities in height and weight between the
sexes at ages 7 and 8, this pre-pubertal group of
children were already showing body composition
differences commensurate with their gender. While
BMI was similar between the sexes, values for percentage body fat (calculated from both skinfolds and
BIA) suggested that the girls were fatter than the boys.
Similar observations about pre-pubertal differences in
fat and FFM have been made by others.14,15 Pietrobelli
et al 15 detected differences in fatness between boys
and girls (P < 0.05) using BMI as a proxy for body
fatness. We found no such difference using this
method in our sample.
A possible explanation for the apparent lack of
agreement between percentage body fat and BMI in
our study could be that boys had more FFM but less
body fat than girls for a similar weight. Thus, differences in the proportion of these components might be
masked by BMI, which is dependent on weight and
height. Bandini et al 16 have also reported a low
accuracy of BMI compared with skinfolds when
estimating fatness in pre-pubertal children.
When the changes in anthropometry and body
composition occurring between baseline and followup were examined, all measures increased except
percentage FFM in both sexes and percentage body
fat from skinfolds in boys. This may have been
Body composition in children
CHS Ruxton et al
1280
because boys were accumulating truncal rather than
peripheral fat mass, since the estimate of fat mass
from BIA did increase signi®cantly over the 12
months. However, it is acknowledged that the BIA
longitudinal sample was small and this may have
contributed bias. The only way of properly assessing
differences in fat compartments over time is by using
a multi-compartmental model, which would have been
inappropriate for a ®eld study design such as ours.
However, there are bene®ts of our two-compartment
®eld study over that of a technically superior multicompartmental study. Sample sizes can be larger,
ethical approval is more likely to be given and a
more representative group of children can be studied.
The change in percentage FFM in both boys and
girls over time was intriguing because the value went
down instead of up. It seemed that FFM as a proportion of body weight was being lost in favour of a gain
in fat mass. This is a worrying trend and could
represent insuf®cient exercise in this age group.
Girls appeared to be gaining percentage body fat
and losing percentage FFM at a greater rate compared
with boys, but the difference failed to reach signi®cance. This may have been because variation in
percentage body fat was too broad within our
sample size, and it would be interesting to repeat the
gender comparison in a larger group.
The primary purpose of this study was to provide
normative data on body composition, and longitudinal
body composition changes in a sample representative of
Edinburgh 7 ± 8 y-olds. The choice of the skinfold
thickness method to estimate body composition might
be considered unusual and needs to be justi®ed. The
advantages of skinfold thicknesses for our study were
deemed to be twofold. First, in ®eld-based community
studies of this type, `®eld' methods of estimating body
composition (for example skinfolds, BIA) are the only
practical options.2 Second, we have previously shown
that the Slaughter prediction equation,10 derived originally from a multi-component model, provided unbiased
estimates of body fatness relative to hydrodensitometry
in Scottish pre-pubertal children.11 The use of skinfolds
is known to be subject to technical error, but this was
minimized by the use of a single trained observer,
although it is worth considering that such errors are
inherent even in reference methods such as hydrodensitometry.1 Errors in prediction equations to estimate
body composition from skinfolds have been subjected
to detailed considerations of acceptability1 and the
equations used here produced errors which were
within the limits of acceptability.2,11
The comparison with the reference child,3 based on
theoretical considerations and data collected in the
70s, revealed that our sample was taller, heavier and
fatter. Indeed, a study of growth trends in over 7000
British children between 1972 and 19946 revealed a
similar ®nding. The striking differences arose when
FFM and fat mass were expressed as a percentage of
weight. Our sample demonstrated proportionally more
fat and less FFM at both ages compared with the
reference child. When growth rates were examined,
our sample was gaining height at a similar rate to the
reference child, but putting on proportionally more
weight and body fat and failing to gain suf®cient
FFM. This lends support to a worrying trend that
our rapidly growing modern children are getting
rapidly fatter as well.
Two implications arise from this ®nding. If the
reference child is no longer an appropriate representation of normal body composition, current assumptions
about the nature of FFM in children may need to be
revised since many of these are based on Fomon's
data. Since age-speci®c assumptions about the composition of FFM underpin many body composition
methods, an investigation into the validity of the
constants used in these methods is warranted. Such
an approach would require empirical data based on
multi-component models Ð an option that was not
available at the time of the construction of Fomon's
reference child.
A second implication of the present study's ®ndings
relates to the use of body composition standards in
clinical and research situations to estimate the prevalence of childhood obesity. While it is possibly
acceptable that a new reference would take into
account secular changes in height and weight, any
upward revision of the level of `normal' fatness might
not be appropriate in a public health context where a
key aim is to identify and treat children whose body
fat levels are unacceptably high. A full debate is
needed on how this issue should be resolved.
To conclude, this study demonstrates that modern
children have outgrown the Fomon standard for the
`reference child' and are now taller, heavier and
fatter than expected. This may have implications for
paediatric body composition methodology and the use
of such standards for public health purposes.
Acknowledgements
The project was ®nancially supported by a Queen
Margaret College studentship (CHSR) and contributions from the Kellogg Company of Great Britain and
the Scottish Dairy Council. The assistance of Dr
Angelo Pietrobelli in reviewing this manuscript is
gratefully acknowledged.
References
1 Lohman TG. Advances in body composition assessment.
Monograph no. 3. Human Kinetics Publishers: Champaign,
IL, 1993.
2 Reilly JJ. Assessment of body composition in infants and
children. Nutrition 1998; 14: 821 ± 825.
3 Fomon SJ, Haschke F, Ziegler EE, Nelson SE. Body composition of reference children from birth to age 10 years. Am J Clin
Nutr 1982; 35: 1169 ± 1175.
4 Hamill PVV, Drizd TA, Johnson CL, Reed RB, Roche AF,
Moore WM. Physical growth: NCHS percentiles. Am J Clin
Nutr 1979; 32: 607 ± 629.
5 Troiano RP, Flegal KM, Kuczmarski RJ, Campbell SM,
Johnson CL. Overweight prevalence and trends for children
and adolescents. Arch Pediatr Adolesc Med 1995; 149:
1085 ± 1091.
Body composition in children
CHS Ruxton et al
6 Hughes JM, Li L, Chinn S, Rona RJ. Trends in growth in England
and Scotland, 1972 to 1994. Arch Dis Child 1997; 76: 182 ± 189.
7 Prentice A, Jebb S. Obesity in Britain: gluttony or sloth? Br
Med J 1995; 311: 437 ± 439.
8 Ruxton CHS, Kirk TR, Belton NR, Holmes MAM. The
in¯uence of social class on nutrient intakes and dietary
patterns in a sample of Edinburgh school children. Int J
Food Sci Nutr 1996; 47: 341 ± 349.
9 Tanner JM, Whitehouse RH, Takaishi M. Standards from birth
to maturity for height, weight, height velocity, and weight
velocity: British children, 1965. Arch Dis Child 1966; 41:
454 ± 471, 613 ± 635.
10 Slaughter MH, Lohman TG, Boileau RA et al. Skinfold
equations for estimation of body fatness in children and
youth. Hum Biol 1988; 60: 709 ± 723.
11 Reilly JJ, Wilson JJ, Durnin JVGA. Determination of body
composition from skinfold thickness: a validation study. Arch
Dis Child 1995; 73: 305 ± 310.
12 Houtkooper LB, Going SB, Lohman TG, Roche AF, Van Loan
M. Bio-electrical impedance estimation of fat-free body mass
in children and youth: a cross-validation study. J Appl Physiol
1992; 72: 366 ± 373.
13 Reilly JJ, Wilson J, McColl JH, Carmichael M, Durnin JVGA.
Ability of bioelectric impedance to predict fat-free mass in
prepubertal children. Pediatr Res 1996; 39: 176 ± 179.
14 Taylor RW, Gold E, Manning P, Goulding A. Gender differences in body fat content are present well before puberty. Int J
Obes 1997; 21: 1082 ± 1084.
15 Pietrobelli A, Faith MS, Allison DB, Gallagher D, Chiumello
G, Heyms®eld SB. Body mass index as a measure of adiposity
among children and adolescents: a validation study. J Pediatr
1998; 132: 204 ± 210.
16 Bandini LG, Vu DM, Must A, Dietz WH. Body fatness and
bioelectrical impedance in non-obese pre-menarcheal girls:
comparison to anthropometry and evaluation of predictive
equations. Eur J Clin Nutr 1997; 51: 673 ± 677.
1281